1. Cholinergic Sensorimotor Integration Regulates Olfactory Steering
He Liu, Wenxing Yang, Taihong Wu, Fengyun Duan, Edward Soucy, Xin Jin, Yun Zhang 
Neuron 
Volume 97, Issue 2, Pages e3 (January 2018)
DOI: /j.neuron
Copyright © 2017 Elsevier Inc. Terms and Conditions

2. Figure 1
The Interneuron RIA Is Involved in Olfactory Steering in C. elegans
(A) Schematics for the navigation index during steering toward an attractive odorant. The end point of steering is on the edge of the drop of the odorant source.
(B and C) Navigation index decreases (B) and traveling distance before reaching isoamyl alcohol (IAA) increases (C) when animals steer toward decreasing concentrations of IAA. n = 30 (9.2 × 10−3 M), 29 (9.2 × 10−4 M), 34 (9.2 × 10−5 M), and 23 (9.2 × 10−6 M) animals, one-way ANOVA.
(D) Schematics showing the sensory and motor circuits that anatomically converge on the axon of RIA. Arrows denote synapses (White et al., 1986); the RIA soma, nrV, nrD, and loop axonal domains are highlighted with colors; and the synaptic outputs of nrV and nrD are highlighted with blue and red, respectively. The characterized neurotransmitters in the circuits are shown. Glu, glutamate; Ach, acetylcholine.
(E and F) Selectively blocking the synaptic release of RIA by expressing the tetanus toxin (TeTx) reduces the navigation index (E) and increases the total traveling distance (F) when steering toward IAA (9.2 × 10−3 M). n = 47 wild-type, n = 53 RIA::Tetx transgenic animals, Student’s t test.
(G) Expressing the tetanus toxin in RIA does not alter the average duration of one head undulation. n = 16 wild-type, n = 22 RIA::TeTx transgenic animals, Student’s t test.
∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05; N.S., not significant. The boxplots show the first quartile, median, and third quartile; the whiskers extend to the data points that are equal to or less than 1.5 IQR (interquartile range) away from the first or the third quartile; dots are outliers.
See also Figure S1.
Neuron  , e3DOI: ( /j.neuron )
Copyright © 2017 Elsevier Inc. Terms and Conditions

3. Figure 2
RIA Responds to the Pulses of Olfactory Stimulation as Rapidly as the Head Undulates
(A) Sample traces for the GCaMP3 signals of the nrV and the nrD axonal domains of RIA and the head bending angles of one worm that is stimulated with 0.5 Hz pulses of isoamyl alcohol (IAA, 9.2 × 10−3 M) in a microfluidic chip (left panel) and the average GCaMP3 signals per stimulation cycle for the trace in the left panel (right panel; solid lines represent mean; shades represent SEM). Fbase is the mean intensity of the axonal signals in a few frames of low-activity state; ΔF indicates F − Fbase.
(B) Cross-correlation between the RIA axonal GCaMP3 signals and the head bendings in a microfluidic chip. Ventral direction is positive, solid lines represent mean, and shades represent SEM, n = 8 animals.
(C and D) Samples of the Fourier transforms of the GCaMP3 signals of nrV (C; n = 15 animals), nrD (C; n = 15 animals), or loop (D; n = 7 animals) domains of RIA show the peak at 0.5 Hz, which is the frequency of IAA stimulation. Solid lines represent mean; shades represent SEM.
(E) Genetic ablation of AWC neuron (AWC::Caspase, n = 12 animals) or mutating eat-4 (n = 8 animals) significantly reduces the amplitudes of the 0.5 Hz peak identified by the Fourier transform of the nrV and the nrD GCaMP3 signals in RIA in comparison with wild-type (n = 12 animals). One-way ANOVA; the boxplots show the first quartile, median, and third quartile; the whiskers extend to the data points that are equal to or less than 1.5 IQR away from the first or the third quartile; dots are outliers.
(F and G) Mean traces of the GCaMP6 signals (left panel, F) in the interneurons AIY (n = 15 animals) and AIB (n = 14 animals) in response to 0.5 Hz pulses of IAA (9.2 × 10−3 M) and the average GCaMP6 signals per stimulation cycle for the traces in the left panel (right panel, F; n, number of worms), and the Fourier transforms of the GCaMP6 signals (G). Solid lines represent mean; shades represent SEM. Fbase is the mean intensity during the 2 s window before the first onset of IAA; ΔF indicates F − Fbase.
(H) Mutating ttx-3 (n = 11 animals) or blocking the synaptic release from AIY (AIY::TeTx, n = 11 animals) significantly reduces the amplitude of the 0.5 Hz peak identified by the Fourier transform of the GCaMP3 signals of nrV and nrD evoked by 0.5 Hz IAA pulses in comparison with wild-type (n = 15 animals). One-way ANOVA; the boxplots show the first quartile, median, and third quartile; the whiskers extend to the data points that are equal to or less than 1.5 IQR away from the first or the third quartile; dots are outliers.
∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < See also Figure S2 and Movies S1 and S2.
Neuron  , e3DOI: ( /j.neuron )
Copyright © 2017 Elsevier Inc. Terms and Conditions

4. Figure 3
Cholinergic Neurotransmission Regulates the Sensory-Evoked and Motor-Encoding Activities in the RIA Axonal Domains
(A) Sample traces for the GCaMP3 signals of the nrV and the nrD axonal domains of RIA and for the head bendings in a gar-3(gk305) mutant animal stimulated with 0.5 Hz pulses of IAA (9.2 × 10−3 M) in a microfluidic chip. Ventral direction is positive, Fbase is the mean intensity of the axonal signals in a few frames of low-activity state, and ΔF indicates F − Fbase.
(B) When stimulated with 0.5 Hz pulses of IAA, mutating gar-3 (n = 8 animals) significantly decreases the peak correlations between the nrV and the nrD GCaMP3 signals with the head bendings compared with wild-type (n = 8 animals).
(C) The amplitudes of the 0.5 Hz peak identified by the Fourier transform of the nrV or the nrD GCaMP3 signals are comparable in wild-type (n = 15 animals), the transgenic animals expressing a wild-type gar-3 cDNA in the gar-3(gk305) mutant background (n = 10 animals), and the gar-3 mutants (n = 9 animals) in response to 0.5 Hz IAA pulses.
(D) Sample traces for RIA axonal GCaMP5 signals and the head bendings in an acc-2(ok2216) mutant animal stimulated with 0.5 Hz pulses of IAA (9.2 × 10−3 M) in a microfluidic chip. Ventral direction is positive, Fbase is the mean intensity of the axonal signals in a few frames of low-activity state, and ΔF indicates F − Fbase.
(E) When stimulated with 0.5 Hz pulses of IAA, mutating acc-2 (n = 14 animals) significantly decreases the amplitude of the 0.5 Hz peak identified by the Fourier transform of the nrV and the nrD GCaMP5 signals compared with wild-type (n = 14 animals), and the defect was rescued by expressing a wild-type acc-2 cDNA in RIA (n = 11 animals).
(F) The acc-2 mutation (n = 7 animals) does not alter the peak correlations between the nrV and the nrD GCaMP5 signals with the head bendings in comparison with wild-type (n = 8 animals) when stimulated with 0.5 Hz pulses of IAA.
Student’s t test (B and F); one-way ANOVA (C and E). ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05; N.S., not significant. The boxplots show the first quartile, median, and third quartile; the whiskers extend to the data points that are equal to or less than 1.5 IQR away from the first or the third quartile; dots are outliers.
See also Figure S3.
Neuron  , e3DOI: ( /j.neuron )
Copyright © 2017 Elsevier Inc. Terms and Conditions

5. Figure 4
The Sensory-Evoked Activity of RIA Suppresses the Motor-Encoding Activity of RIA in a Microfluidic Device
(A and B) Sample frames for the GCaMP3 signals of nrV, nrD, and loop domains of RIA during a ventral head bending (A) or a dorsal head bending (B). The frames are pseudo-colored (ImageJ) to better demonstrate GCaMP3 signals. A, anterior; D, dorsal. Scale bar, 50 μm. The worm body is inside of a channel with the head outside.
(C) Sample plot showing the linear regression fit of the GCaMP3 signal of nrV in one worm as a function of the ventral or the dorsal head bending angles. Ventral direction is positive, each dot indicates the nrV GCaMP3 signal and the head bending angle in one frame, and the plot represents the results from one movie.
(D and E) Compared with the linear fit of the nrV GCaMP3 signals and the dorsal head bending angles, the linear regression fit of the nrV GCaMP3 signals and the ventral head bending angles generates steeper slopes (D) and bigger coefficients of determination (E). n = 6 animals, Student’s t test.
(F) Sample plot showing the linear regression fit of the GCaMP3 signal of nrD in one worm as a function of the ventral or the dorsal head bending angles. Ventral direction is positive and each dot indicates the GCaMP3 signal of nrD and the head bending angle in one movie frame; the plot represents the results from one movie.
(G and H) Compared with the linear fit of the nrD GCaMP3 signals and the ventral head bending angles, the linear regression fit of the nrD GCaMP3 signals and the dorsal head bending angels generates steeper slopes (G) and bigger coefficients of determination (H). n = 6 animals, Student’s t test.
(I–L) In comparison with the buffer pulses (n = 22 animals), IAA pulses (9.2 × 10−3 M, n = 29 animals) significantly reduce the steepness of the slopes for the linear fit of the nrV GCaMP3 signals and the ventral head bending angles (I) and for the linear fit of the nrD GCaMP3 signals and the dorsal head bending angles (K) without altering the coefficients of determination (J and L). Student’s t test.
∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05; N.S., not significant. The boxplots show the first quartile, median, and third quartile; the whiskers extend to the data points that are equal to or less than 1.5 IQR away from the first or the third quartile; dots are outliers. Fbase is the mean intensity of the axonal signals in a few frames of low-activity state; ΔF indicates F − Fbase.
Neuron  , e3DOI: ( /j.neuron )
Copyright © 2017 Elsevier Inc. Terms and Conditions

6. Figure 5
Cholinergic Neurotransmission Regulates Sensorimotor Integration in RIA during Olfactory Steering toward IAA
(A) Sample traces for the GCaMP3 signals of the nrV and the nrD axonal domains of RIA and the amplitudes of head undulations in a freely moving animal. Ventral direction is positive, Fmin is the minimal intensity for the movie, and ΔF indicates F − Fmin.
(B) The GCaMP3 signals of nrV and nrD correlate with the ventral and dorsal head undulations, respectively, in freely moving animals. n = 27 animals.
(C) The GCaMP3 signals of nrV and nrD normalized with the mCherry signals of the same domains correlate with the ventral and the dorsal head undulations, respectively, in freely moving animals. n = 28 animals.
(D) The GFP signals of nrV and nrD do not correlate with the head undulations in freely moving animals. n = 9 animals.
In (B)–(D), ventral direction is positive, solid lines represent mean, and shades represent SEM.
(E) Schematics showing the procedure of in vivo calcium imaging in freely moving animals first in the absence of any odorant and then in the presence of IAA.
(F) The GCaMP3 signals of nrV and nrD are significantly smaller during olfactory steering toward a drop of IAA (9.2 × 10−3 M) than during freely moving in the absence of an odorant. n = 33 animals.
(G) The sensory-evoked suppression of the GCaMP3 signals in the RIA axonal domains in steering animals is abolished in eat-4(ky5) mutants. n = 22 animals.
(H and I) Mutating gar-3 abolishes the correlation of the RIA axonal GCaMP3 signals with the head undulations while maintaining the sensory-evoked suppression of the axonal activity (H; n = 17 animals); expressing a wild-type gar-3 cDNA selectively in RIA rescues the defect (I; n = 25 animals).
(J and K) Mutating acc-2 abolishes the sensory-evoked suppression of the RIA axonal GCaMP5 signals (J; n = 24 animals); expressing a wild-type acc-2 cDNA selectively in RIA rescues the defect (K; n = 26 animals).
In (F)–(K), Fmin is the minimal intensity of GCaMP3 or GCaMP5 signals for each worm; ΔF indicates F − Fmin. The GCaMP signals of nrV and nrD over each complete head undulation (for nrV, from the maximal dorsal [Max-D] position to the maximal ventral [Max-V] positon and back to the maximal dorsal [Max-D] position; for nrD, from the maximal ventral [Max-V] position to the maximal dorsal [Max-D] positon and back to the maximal ventral [Max-V] position) are measured and the averages for all the head undulations made by each worm are generated, which are used to generate mean and SEM, represented by solid lines and shades, respectively, for multiple worms. Kolmogorov-Smirnov test (MATLAB) is used to compare the mean values, ∗∗∗p < 0.001; N.S., not significant.
See also Figure S4 and Movie S3.
Neuron  , e3DOI: ( /j.neuron )
Copyright © 2017 Elsevier Inc. Terms and Conditions

7. Figure 6
The Cholinergic Sensorimotor Integration in RIA Decodes the Spatial Information of the Odorant to Bias the Steering Trajectory
(A) Schematics showing the sampling head movements when a worm steers toward isoamyl alcohol (IAA) located on the ventral or the dorsal side of the animal.
(B) The GCaMP3 signal of nrV is significantly weaker than that of nrD when animals steer toward IAA located on the ventral side of the animals (n = 18 animals), and the GCaMP3 signal of nrD is significantly weaker than that of nrV when animals steer toward IAA located on the dorsal side (n = 15 animals).
(C) The mutation in eat-4(ky5) abolishes the asymmetry between the RIA axonal GCaMP3 signals during steering toward IAA regardless of the location of IAA. n = 11 animals each condition.
(D) A sample image of an RIA neuron expressing miniSOG(103L)-VAMP2-mCherry. RIA soma and the axonal domains are denoted. A, anterior; D, dorsal. Scale bar, 10 μm.
(E) Sample locomotory traces generated by animals in which the synaptic release of the nrV or the nrD domain of RIA is selectively blocked with a blue-light illumination of a miniSOG(103L)-VAMP2-mCherry fusion protein expressed in RIA and a sample trace of a control animal that expresses miniSOG(103L)-mCherry in RIA. Ventral or dorsal side of the animal is denoted and arrows denote movement directions.
(F) Mean angle of the head undulations generated by animals in which the synaptic release of nrV (n = 10 and 7 animals for experiment and control, respectively) or nrD (n = 7 and 6 animals for experiment and control, respectively) domain of RIA is selectively blocked. Ventral direction is positive. Student’s t test, ∗p < The boxplots show the first quartile, median, and third quartile; the whiskers extend to the data points that are equal to or less than 1.5 IQR away from the first or the third quartile; dots are outliers.
(G and H) Mutating gar-3 abolishes the asymmetry between the GCaMP3 signals of nrV and nrD during olfactory steering toward IAA regardless of the location of IAA (G; n = 9 animals IAA on ventral side; n = 8 animals IAA on dorsal side), and expressing a wild-type gar-3 cDNA selectively in RIA rescues the defect (H; n = 14 animals IAA on ventral side; n = 11 animals IAA on dorsal side).
(I and J) Mutating acc-2 abolishes the asymmetry between the GCaMP5 signals of nrV and nrD during olfactory steering toward IAA regardless of the location of IAA (I; n = 13 animals IAA on ventral side; n = 11 animals IAA on dorsal side), and the defect is rescued by expressing a wild-type acc-2 cDNA selectively in RIA (J; n = 15 animals IAA on ventral side; n = 11 animals IAA on dorsal side).
In (B), (C), and (G)–(J), Fmin is the minimal intensity of GCaMP3 or GCaMP5 for each worm; ΔF indicates F − Fmin. The GCaMP signals of nrV and nrD over each complete head undulation (for nrV, from the maximal dorsal position to the maximal ventral positon and back to the maximal dorsal position; for nrD, from the maximal ventral position to the maximal dorsal positon and back to the maximal ventral position) are measured and the averages for all the head undulations made by each worm are generated, which are used to generate mean and SEM, represented by solid lines and shades, respectively, for multiple worms. Kolmogorov-Smirnov test (MATLAB) is used to compare the mean values, ∗∗∗p < 0.001, ∗∗p < 0.01; N.S., not significant.
(K–N) The navigation index for gar-3(gk305) (K) or acc-2(ok2216) (M) mutants is significantly smaller than that for wild-type when steering toward IAA, and the defects are rescued by expressing a wild-type gar-3 cDNA or a wild-type acc-2 cDNA selectively in RIA in the respective mutant animals (K and M); meanwhile, the distance traveled by gar-3(gk305) animals (L) or acc-2(ok2216) animals (N) before reaching IAA is significantly longer than wild-type, and the defects are rescued by expressing a wild-type gar-3 cDNA or a wild-type acc-2 cDNA selectively in RIA in the respective mutant animals (L and N). n = 61 (wild-type), 59 (gar-3 mutant), and 40 (gar-3; RIA::gar-3 transgenic) animals (K and L); n = 35 (wild-type), 43 (acc-2 mutant), and 41 (acc-2; RIA::acc-2 transgenic) animals (M and N). One-way ANOVA, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05; N.S., not significant. The boxplots show the first quartile, median, and third quartile; the whiskers extend to the data points that are equal to or less than 1.5 IQR away from the first or the third quartile; dots are outliers.
See also Figure S4.
Neuron  , e3DOI: ( /j.neuron )
Copyright © 2017 Elsevier Inc. Terms and Conditions

8. Figure 7
Aversive Olfactory Training Modulates Sensorimotor Integration in RIA to Generate Experience-Specific Changes in Olfactory Steering
(A–D) When stimulated with 0.5 Hz pulses of the supernatant of the culture of P. aeruginosa PA14, animals trained with PA14 display a decreased amplitude in the 0.5 Hz peak identified by the Fourier transform of the GCaMP3 signals and decreased average GCaMP3 signals per stimulation cycle of the nrV (A) and the nrD (B) axonal domains of RIA (A and B; n = 9 naive animals, n = 11 trained animals); however, when stimulated with 0.5 Hz pulses of the supernatant of E. coli OP50, animals trained with PA14 display only decreased average axonal GCaMP3 signals per stimulation cycle without altering the amplitude of the 0.5 Hz peak identified by the Fourier transform of the GCaMP3 signals (C and D; n = 13 naive animals, n = 15 trained animals). Left panels show the average traces of the Fourier transforms of the GCaMP3 signals of multiple worms and the right panels show the average GCaMP3 signals per stimulation cycle of multiple worms. n, number of worms. Solid lines represent mean; shades represent SEM. Fbase is the mean intensity of the axonal signals in a few frames of low-activity state; ΔF indicates F − Fbase. Kolmogorov-Smirnov test (MATLAB) is used to compare the mean values of ΔF/Fbase %, ∗∗∗p < 0.001, ∗∗p < 0.01.
(E and F) Quantitation of the amplitudes of the 0.5 Hz peaks identified by the Fourier transform of the RIA axonal GCaMP3 signals shown in (A) and (B) (E; n = 9 naive animals, n = 11 trained animals) and (C) and (D) (F; n = 13 naive animals, n = 15 trained animals).
(G–J) Animals trained with PA14 display a decreased navigation index (G) and a longer traveling distance (H) when steering toward the supernatant of PA14 in comparison with naive animals (G and H; n = 29 naive animals, n = 25 trained animals), but display a navigation index (I) and a traveling distance (J) that are similar to those in naive animals when steering toward the supernatant of OP50 (I and J; n = 41 naive animals, n = 46 trained animals).
In (E)–(J), Student’s t test, ∗∗∗p < 0.001, ∗∗p < 0.01; N.S., not significant. The boxplots show the first quartile, median, and third quartile; the whiskers extend to the data points that are equal to or less than 1.5 IQR away from the first or the third quartile; dots are outliers.
See also Figure S5.
Neuron  , e3DOI: ( /j.neuron )
Copyright © 2017 Elsevier Inc. Terms and Conditions